AI allows hospital to maintain MRI productivity with 1 fewer scanner
Artificial intelligence is allowing one hospital to maintain its MRI productivity even with one fewer scanner.
Magnetic resonance imaging utilization is steadily growing, experts note, driven by expanding clinical indications, legal risk mitigation and an aging population. However, growth can bring substantial financial challenges, including high procurement and operating costs, researchers write in the European Journal of Radiology.
University of Oulu, Finland, is detailing how its radiology department was able to tackle these challenges without adding more imaging machines. It’s doing so with the help of deep learning reconstruction technology, or DLR, allowing the hospital to deliver scans faster with comparable image quality.
“[Deep learning reconstruction] deployment was associated with improved MRI suite productivity, enabling nearly pre-DLR throughput despite operating with one fewer scanner,” Mikael Brix, PhD, a physicist and postdoctoral research at the University of Oulu, and co-authors wrote Feb. 23. “The results demonstrate the utility of DLR in improving MRI productivity and support the predictive accuracy of simulation-based health technology assessment.”
Their research traces to late 2024, when the hospital combined its pediatric and emergency imaging departments into one, relocating into a new building. The move reduced joint capacity for the department, forcing radiologists and techs to work with fewer scanners. Around that same time, Oulu introduced use of the key DLR software, which is manufactured by Siemens Healthineers. Brix and colleagues based their study on a retrospective analysis of MRI scanner log data and image assessments conducted as part of routine quality assurance.
The study covered 10 months in 2023 before implementing deep learning and the same 10 months in 2025 after DLR and with one fewer MRI machine. Optimized scanners demonstrated a total reduction of between 5 minutes to 11 minutes (11.5%–27.2%) in sequence duration. They also saw a 5- to nearly 11-minute reduction in total exam times (9.5%–21.2%). Despite operating with one fewer scanner in 2025, the average hourly throughput of the entire MRI fleet dropped by only 6.4%, “indicating improved productivity per scanner.”
“Notably, this throughput was achieved while DLR deployment and protocol optimization were still in progress, underscoring the substantial productivity benefit even at a partial implementation stage,” Brix and colleagues noted.
However, “unpredictable performance” may limit the applicability of DLR, the study found. This was particularly relevant in neuroimaging, where the research team discovered artifacts in T2-weighted scan sequences and reduced quality in contrast-enhanced studies. This also underlines the need for “rigorous quality assurance” when implementing deep learning in clinical practice.
“Future research on new radiological technologies should integrate financial and productivity indicators alongside diagnostic value to give decision-makers a more comprehensive understanding of the overall impact,” the authors advised.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.